Dynamic Complexity of Expansion

TL;DR

This paper extends the understanding of dynamic complexity by showing that spectral expansion properties of bounded degree graphs can be maintained efficiently under batch updates using linear algebra techniques.

Contribution

It demonstrates that the spectral expansion gap in bounded degree graphs can be preserved in the class ynACz under batch edge modifications, extending previous work with linear algebra methods.

Findings

Spectral expansion gap maintained under batch updates.

Linear algebra techniques applied to dynamic graph properties.

Maintains transition matrix powers in ynACz.

Abstract

Dynamic Complexity was introduced by Immerman and Patnaik \cite{PatnaikImmerman97} (see also \cite{DongST95}). It has seen a resurgence of interest in the recent past, see \cite{DattaHK14,ZeumeS15,MunozVZ16,BouyerJ17,Zeume17,DKMSZ18,DMVZ18,BarceloRZ18,DMSVZ19,SchmidtSVZK20,DKMTVZ20} for some representative examples. Use of linear algebra has been a notable feature of some of these papers. We extend this theme to show that the gap version of spectral expansion in bounded degree graphs can be maintained in the class $\DynACz$ (also known as $\dynfo$, for domain independent queries) under batch changes (insertions and deletions) of $O(\frac{\log{n}}{\log{\log{n}}})$ many edges. The spectral graph theoretic material of this work is based on the paper by Kale-Seshadri \cite{KaleS11}. Our primary technical contribution is to maintain up to logarithmic powers of the transition matrix of a bounded degree undirected graph in $\DynACz$.